1
0
mirror of https://github.com/hashcat/hashcat.git synced 2024-12-29 01:48:41 +00:00
hashcat/OpenCL/m14900_a3-optimized.cl

360 lines
16 KiB
Common Lisp

/**
* Author......: See docs/credits.txt
* License.....: MIT
*/
//too much register pressure
//#define NEW_SIMD_CODE
#include "inc_vendor.cl"
#include "inc_hash_constants.h"
#include "inc_hash_functions.cl"
#include "inc_types.cl"
#include "inc_common.cl"
#include "inc_simd.cl"
__constant u8a c_ftable[256] =
{
0xa3, 0xd7, 0x09, 0x83, 0xf8, 0x48, 0xf6, 0xf4,
0xb3, 0x21, 0x15, 0x78, 0x99, 0xb1, 0xaf, 0xf9,
0xe7, 0x2d, 0x4d, 0x8a, 0xce, 0x4c, 0xca, 0x2e,
0x52, 0x95, 0xd9, 0x1e, 0x4e, 0x38, 0x44, 0x28,
0x0a, 0xdf, 0x02, 0xa0, 0x17, 0xf1, 0x60, 0x68,
0x12, 0xb7, 0x7a, 0xc3, 0xe9, 0xfa, 0x3d, 0x53,
0x96, 0x84, 0x6b, 0xba, 0xf2, 0x63, 0x9a, 0x19,
0x7c, 0xae, 0xe5, 0xf5, 0xf7, 0x16, 0x6a, 0xa2,
0x39, 0xb6, 0x7b, 0x0f, 0xc1, 0x93, 0x81, 0x1b,
0xee, 0xb4, 0x1a, 0xea, 0xd0, 0x91, 0x2f, 0xb8,
0x55, 0xb9, 0xda, 0x85, 0x3f, 0x41, 0xbf, 0xe0,
0x5a, 0x58, 0x80, 0x5f, 0x66, 0x0b, 0xd8, 0x90,
0x35, 0xd5, 0xc0, 0xa7, 0x33, 0x06, 0x65, 0x69,
0x45, 0x00, 0x94, 0x56, 0x6d, 0x98, 0x9b, 0x76,
0x97, 0xfc, 0xb2, 0xc2, 0xb0, 0xfe, 0xdb, 0x20,
0xe1, 0xeb, 0xd6, 0xe4, 0xdd, 0x47, 0x4a, 0x1d,
0x42, 0xed, 0x9e, 0x6e, 0x49, 0x3c, 0xcd, 0x43,
0x27, 0xd2, 0x07, 0xd4, 0xde, 0xc7, 0x67, 0x18,
0x89, 0xcb, 0x30, 0x1f, 0x8d, 0xc6, 0x8f, 0xaa,
0xc8, 0x74, 0xdc, 0xc9, 0x5d, 0x5c, 0x31, 0xa4,
0x70, 0x88, 0x61, 0x2c, 0x9f, 0x0d, 0x2b, 0x87,
0x50, 0x82, 0x54, 0x64, 0x26, 0x7d, 0x03, 0x40,
0x34, 0x4b, 0x1c, 0x73, 0xd1, 0xc4, 0xfd, 0x3b,
0xcc, 0xfb, 0x7f, 0xab, 0xe6, 0x3e, 0x5b, 0xa5,
0xad, 0x04, 0x23, 0x9c, 0x14, 0x51, 0x22, 0xf0,
0x29, 0x79, 0x71, 0x7e, 0xff, 0x8c, 0x0e, 0xe2,
0x0c, 0xef, 0xbc, 0x72, 0x75, 0x6f, 0x37, 0xa1,
0xec, 0xd3, 0x8e, 0x62, 0x8b, 0x86, 0x10, 0xe8,
0x08, 0x77, 0x11, 0xbe, 0x92, 0x4f, 0x24, 0xc5,
0x32, 0x36, 0x9d, 0xcf, 0xf3, 0xa6, 0xbb, 0xac,
0x5e, 0x6c, 0xa9, 0x13, 0x57, 0x25, 0xb5, 0xe3,
0xbd, 0xa8, 0x3a, 0x01, 0x05, 0x59, 0x2a, 0x46
};
void g (__local u8 *s_ftable, const u32 *key, const int k, const u32 *wx, u32 *out)
{
const u32 g1 = wx[1];
const u32 g2 = wx[0];
const u32 g3 = s_ftable[g2 ^ key[(4 * k + 0) % 10]] ^ g1;
const u32 g4 = s_ftable[g3 ^ key[(4 * k + 1) % 10]] ^ g2;
const u32 g5 = s_ftable[g4 ^ key[(4 * k + 2) % 10]] ^ g3;
const u32 g6 = s_ftable[g5 ^ key[(4 * k + 3) % 10]] ^ g4;
out[0] = g6;
out[1] = g5;
}
u32 skip32 (__local u8 *s_ftable, const u32 KP, const u32 *key)
{
u32 wl[2];
u32 wr[2];
wl[0] = (KP >> 8) & 0xff;
wl[1] = (KP >> 0) & 0xff;
wr[0] = (KP >> 24) & 0xff;
wr[1] = (KP >> 16) & 0xff;
for (u32 i = 0; i < 12; i++)
{
const u32 k0 = (i * 2) + 0;
const u32 k1 = (i * 2) + 1;
u32 tmp[2];
g (s_ftable, key, k0, wl, tmp);
tmp[0] ^= k0;
wr[0] ^= tmp[0];
wr[1] ^= tmp[1];
g (s_ftable, key, k1, wr, tmp);
tmp[0] ^= k1;
wl[0] ^= tmp[0];
wl[1] ^= tmp[1];
}
const u32 r = ((wr[1] & 0xff) << 0)
| ((wr[0] & 0xff) << 8)
| ((wl[1] & 0xff) << 16)
| ((wl[0] & 0xff) << 24);
return r;
}
void m14900m (__local u8 *s_ftable, u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 pw_len, __global pw_t *pws, __global const kernel_rule_t *rules_buf, __global const pw_t *combs_buf, __global const bf_t *bfs_buf, __global void *tmps, __global void *hooks, __global const u32 *bitmaps_buf_s1_a, __global const u32 *bitmaps_buf_s1_b, __global const u32 *bitmaps_buf_s1_c, __global const u32 *bitmaps_buf_s1_d, __global const u32 *bitmaps_buf_s2_a, __global const u32 *bitmaps_buf_s2_b, __global const u32 *bitmaps_buf_s2_c, __global const u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global const digest_t *digests_buf, __global u32 *hashes_shown, __global const salt_t *salt_bufs, __global oldoffice01_t *oldoffice01_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV0_buf, __global u32 *d_scryptV1_buf, __global u32 *d_scryptV2_buf, __global u32 *d_scryptV3_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 il_cnt, const u32 digests_cnt, const u32 digests_offset)
{
/**
* modifier
*/
const u64 gid = get_global_id (0);
const u64 lid = get_local_id (0);
/**
* salt
*/
const u32 KP = salt_bufs[salt_pos].salt_buf[0];
/**
* loop
*/
u32 w0l = w0[0];
for (u32 il_pos = 0; il_pos < il_cnt; il_pos += VECT_SIZE)
{
const u32 w0r = ix_create_bft (bfs_buf, il_pos);
const u32 w0lr = w0l | w0r;
u32 key[10];
key[0] = (w0lr >> 0) & 0xff;
key[1] = (w0lr >> 8) & 0xff;
key[2] = (w0lr >> 16) & 0xff;
key[3] = (w0lr >> 24) & 0xff;
key[4] = (w0[1] >> 0) & 0xff;
key[5] = (w0[1] >> 8) & 0xff;
key[6] = (w0[1] >> 16) & 0xff;
key[7] = (w0[1] >> 24) & 0xff;
key[8] = (w0[2] >> 0) & 0xff;
key[9] = (w0[2] >> 8) & 0xff;
const u32 r = skip32 (s_ftable, KP, key);
const u32 z = 0;
COMPARE_M_SIMD (r, z, z, z);
}
}
void m14900s (__local u8 *s_ftable, u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 pw_len, __global pw_t *pws, __global const kernel_rule_t *rules_buf, __global const pw_t *combs_buf, __global const bf_t *bfs_buf, __global void *tmps, __global void *hooks, __global const u32 *bitmaps_buf_s1_a, __global const u32 *bitmaps_buf_s1_b, __global const u32 *bitmaps_buf_s1_c, __global const u32 *bitmaps_buf_s1_d, __global const u32 *bitmaps_buf_s2_a, __global const u32 *bitmaps_buf_s2_b, __global const u32 *bitmaps_buf_s2_c, __global const u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global const digest_t *digests_buf, __global u32 *hashes_shown, __global const salt_t *salt_bufs, __global oldoffice01_t *oldoffice01_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV0_buf, __global u32 *d_scryptV1_buf, __global u32 *d_scryptV2_buf, __global u32 *d_scryptV3_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 il_cnt, const u32 digests_cnt, const u32 digests_offset)
{
/**
* modifier
*/
const u64 gid = get_global_id (0);
const u64 lid = get_local_id (0);
/**
* salt
*/
const u32 KP = salt_bufs[salt_pos].salt_buf[0];
/**
* digest
*/
const u32 search[4] =
{
digests_buf[digests_offset].digest_buf[DGST_R0],
0,
0,
0,
};
/**
* loop
*/
u32 w0l = w0[0];
for (u32 il_pos = 0; il_pos < il_cnt; il_pos += VECT_SIZE)
{
const u32 w0r = ix_create_bft (bfs_buf, il_pos);
const u32 w0lr = w0l | w0r;
u32 key[10];
key[0] = (w0lr >> 0) & 0xff;
key[1] = (w0lr >> 8) & 0xff;
key[2] = (w0lr >> 16) & 0xff;
key[3] = (w0lr >> 24) & 0xff;
key[4] = (w0[1] >> 0) & 0xff;
key[5] = (w0[1] >> 8) & 0xff;
key[6] = (w0[1] >> 16) & 0xff;
key[7] = (w0[1] >> 24) & 0xff;
key[8] = (w0[2] >> 0) & 0xff;
key[9] = (w0[2] >> 8) & 0xff;
const u32 r = skip32 (s_ftable, KP, key);
const u32 z = 0;
COMPARE_S_SIMD (r, z, z, z);
}
}
__kernel void m14900_m04 (__global pw_t *pws, __global const kernel_rule_t *rules_buf, __global const pw_t *combs_buf, __global const bf_t *bfs_buf, __global void *tmps, __global void *hooks, __global const u32 *bitmaps_buf_s1_a, __global const u32 *bitmaps_buf_s1_b, __global const u32 *bitmaps_buf_s1_c, __global const u32 *bitmaps_buf_s1_d, __global const u32 *bitmaps_buf_s2_a, __global const u32 *bitmaps_buf_s2_b, __global const u32 *bitmaps_buf_s2_c, __global const u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global const digest_t *digests_buf, __global u32 *hashes_shown, __global const salt_t *salt_bufs, __global oldoffice01_t *oldoffice01_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV0_buf, __global u32 *d_scryptV1_buf, __global u32 *d_scryptV2_buf, __global u32 *d_scryptV3_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 il_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u64 gid_max)
{
/**
* modifier
*/
const u64 gid = get_global_id (0);
const u64 lid = get_local_id (0);
const u64 lsz = get_local_size (0);
/**
* sbox, kbox
*/
__local u8 s_ftable[256];
for (u32 i = lid; i < 256; i += lsz)
{
s_ftable[i] = c_ftable[i];
}
barrier (CLK_LOCAL_MEM_FENCE);
if (gid >= gid_max) return;
/**
* base
*/
u32 w0[4];
w0[0] = pws[gid].i[ 0];
w0[1] = pws[gid].i[ 1];
w0[2] = pws[gid].i[ 2];
w0[3] = pws[gid].i[ 3];
u32 w1[4];
w1[0] = 0;
w1[1] = 0;
w1[2] = 0;
w1[3] = 0;
u32 w2[4];
w2[0] = 0;
w2[1] = 0;
w2[2] = 0;
w2[3] = 0;
u32 w3[4];
w3[0] = 0;
w3[1] = 0;
w3[2] = 0;
w3[3] = 0;
const u32 pw_len = pws[gid].pw_len;
/**
* main
*/
m14900m (s_ftable, w0, w1, w2, w3, pw_len, pws, rules_buf, combs_buf, bfs_buf, tmps, hooks, bitmaps_buf_s1_a, bitmaps_buf_s1_b, bitmaps_buf_s1_c, bitmaps_buf_s1_d, bitmaps_buf_s2_a, bitmaps_buf_s2_b, bitmaps_buf_s2_c, bitmaps_buf_s2_d, plains_buf, digests_buf, hashes_shown, salt_bufs, oldoffice01_bufs, d_return_buf, d_scryptV0_buf, d_scryptV1_buf, d_scryptV2_buf, d_scryptV3_buf, bitmap_mask, bitmap_shift1, bitmap_shift2, salt_pos, loop_pos, loop_cnt, il_cnt, digests_cnt, digests_offset);
}
__kernel void m14900_m08 (__global pw_t *pws, __global const kernel_rule_t *rules_buf, __global const pw_t *combs_buf, __global const bf_t *bfs_buf, __global void *tmps, __global void *hooks, __global const u32 *bitmaps_buf_s1_a, __global const u32 *bitmaps_buf_s1_b, __global const u32 *bitmaps_buf_s1_c, __global const u32 *bitmaps_buf_s1_d, __global const u32 *bitmaps_buf_s2_a, __global const u32 *bitmaps_buf_s2_b, __global const u32 *bitmaps_buf_s2_c, __global const u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global const digest_t *digests_buf, __global u32 *hashes_shown, __global const salt_t *salt_bufs, __global oldoffice01_t *oldoffice01_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV0_buf, __global u32 *d_scryptV1_buf, __global u32 *d_scryptV2_buf, __global u32 *d_scryptV3_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 il_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u64 gid_max)
{
}
__kernel void m14900_m16 (__global pw_t *pws, __global const kernel_rule_t *rules_buf, __global const pw_t *combs_buf, __global const bf_t *bfs_buf, __global void *tmps, __global void *hooks, __global const u32 *bitmaps_buf_s1_a, __global const u32 *bitmaps_buf_s1_b, __global const u32 *bitmaps_buf_s1_c, __global const u32 *bitmaps_buf_s1_d, __global const u32 *bitmaps_buf_s2_a, __global const u32 *bitmaps_buf_s2_b, __global const u32 *bitmaps_buf_s2_c, __global const u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global const digest_t *digests_buf, __global u32 *hashes_shown, __global const salt_t *salt_bufs, __global oldoffice01_t *oldoffice01_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV0_buf, __global u32 *d_scryptV1_buf, __global u32 *d_scryptV2_buf, __global u32 *d_scryptV3_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 il_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u64 gid_max)
{
}
__kernel void m14900_s04 (__global pw_t *pws, __global const kernel_rule_t *rules_buf, __global const pw_t *combs_buf, __global const bf_t *bfs_buf, __global void *tmps, __global void *hooks, __global const u32 *bitmaps_buf_s1_a, __global const u32 *bitmaps_buf_s1_b, __global const u32 *bitmaps_buf_s1_c, __global const u32 *bitmaps_buf_s1_d, __global const u32 *bitmaps_buf_s2_a, __global const u32 *bitmaps_buf_s2_b, __global const u32 *bitmaps_buf_s2_c, __global const u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global const digest_t *digests_buf, __global u32 *hashes_shown, __global const salt_t *salt_bufs, __global oldoffice01_t *oldoffice01_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV0_buf, __global u32 *d_scryptV1_buf, __global u32 *d_scryptV2_buf, __global u32 *d_scryptV3_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 il_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u64 gid_max)
{
/**
* modifier
*/
const u64 gid = get_global_id (0);
const u64 lid = get_local_id (0);
const u64 lsz = get_local_size (0);
/**
* sbox, kbox
*/
__local u8 s_ftable[256];
for (u32 i = lid; i < 256; i += lsz)
{
s_ftable[i] = c_ftable[i];
}
barrier (CLK_LOCAL_MEM_FENCE);
if (gid >= gid_max) return;
/**
* base
*/
u32 w0[4];
w0[0] = pws[gid].i[ 0];
w0[1] = pws[gid].i[ 1];
w0[2] = pws[gid].i[ 2];
w0[3] = pws[gid].i[ 3];
u32 w1[4];
w1[0] = 0;
w1[1] = 0;
w1[2] = 0;
w1[3] = 0;
u32 w2[4];
w2[0] = 0;
w2[1] = 0;
w2[2] = 0;
w2[3] = 0;
u32 w3[4];
w3[0] = 0;
w3[1] = 0;
w3[2] = 0;
w3[3] = 0;
const u32 pw_len = pws[gid].pw_len;
/**
* main
*/
m14900s (s_ftable, w0, w1, w2, w3, pw_len, pws, rules_buf, combs_buf, bfs_buf, tmps, hooks, bitmaps_buf_s1_a, bitmaps_buf_s1_b, bitmaps_buf_s1_c, bitmaps_buf_s1_d, bitmaps_buf_s2_a, bitmaps_buf_s2_b, bitmaps_buf_s2_c, bitmaps_buf_s2_d, plains_buf, digests_buf, hashes_shown, salt_bufs, oldoffice01_bufs, d_return_buf, d_scryptV0_buf, d_scryptV1_buf, d_scryptV2_buf, d_scryptV3_buf, bitmap_mask, bitmap_shift1, bitmap_shift2, salt_pos, loop_pos, loop_cnt, il_cnt, digests_cnt, digests_offset);
}
__kernel void m14900_s08 (__global pw_t *pws, __global const kernel_rule_t *rules_buf, __global const pw_t *combs_buf, __global const bf_t *bfs_buf, __global void *tmps, __global void *hooks, __global const u32 *bitmaps_buf_s1_a, __global const u32 *bitmaps_buf_s1_b, __global const u32 *bitmaps_buf_s1_c, __global const u32 *bitmaps_buf_s1_d, __global const u32 *bitmaps_buf_s2_a, __global const u32 *bitmaps_buf_s2_b, __global const u32 *bitmaps_buf_s2_c, __global const u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global const digest_t *digests_buf, __global u32 *hashes_shown, __global const salt_t *salt_bufs, __global oldoffice01_t *oldoffice01_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV0_buf, __global u32 *d_scryptV1_buf, __global u32 *d_scryptV2_buf, __global u32 *d_scryptV3_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 il_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u64 gid_max)
{
}
__kernel void m14900_s16 (__global pw_t *pws, __global const kernel_rule_t *rules_buf, __global const pw_t *combs_buf, __global const bf_t *bfs_buf, __global void *tmps, __global void *hooks, __global const u32 *bitmaps_buf_s1_a, __global const u32 *bitmaps_buf_s1_b, __global const u32 *bitmaps_buf_s1_c, __global const u32 *bitmaps_buf_s1_d, __global const u32 *bitmaps_buf_s2_a, __global const u32 *bitmaps_buf_s2_b, __global const u32 *bitmaps_buf_s2_c, __global const u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global const digest_t *digests_buf, __global u32 *hashes_shown, __global const salt_t *salt_bufs, __global oldoffice01_t *oldoffice01_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV0_buf, __global u32 *d_scryptV1_buf, __global u32 *d_scryptV2_buf, __global u32 *d_scryptV3_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 il_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u64 gid_max)
{
}